Simultaneous localization and mapping (SLAM) is often used in the autonomously mobile device, such as a robot, to position. Simultaneous localization and mapping needs lots of sensor, such as global positioning system (GPS), inertial measurement unit (IMU), or odometry, to provide information to position.
For a robot using mecanum wheels or omnidirectional wheel, the speed of the wheel cannot indicate the displacement of the robot and the accumulative error would be increased by the wheel spinning and slipping. When the robot is indoor, the global positioning system is unsuitable, but the inertial measurement unit and the visual odometry is better to detect the displacement of the robot. The visual odometry judges the displacement of the robot by acquiring lots of images of the surrounding environment and comparing the same feature point of the lots of images. However, when the surrounding environment, such as a cement floor without any crack or a white wall, does not have any feature point, the visual odometry cannot judges the displacement of the robot by acquiring enough feature points.
What is needed, therefore, is to provide a self positioning system which can overcome the shortcomings as described above and an autonomously mobile device using the same.